Fluid Cavity - Moving cavity under uniform hydrostatic pressure

[amirbio]

Hi Dear All,

Hope I can have some suggestions to tackle this weird problem.
I have an axisymmetric ball, with CAX8R elements being inflated with hydrostatic fluid elements.
Fluid elements surface is put on the internal surface of the ball (defined by picking the surface of the CAX8R elements).
BC:
Y direction-only allowed for nodes on Y axis.
8 degree of freedom pressurized on the reference node (center of the half circle, cavity reference node), fluid is water.

The internal cavity pressure reaches its expected value at the end of the step. This is very fine and straight forward.
But I see one thing which is rather unrealistic and annoying: The ball moves downward!

My only explanation is that there is something wrong with the normals of the fluid elements, may be not all of them are outward?
But since the surface is automatically defined by abaqus, I dont really know if this is the case and/or how I can tackle it.

Please let me know if I can provide more details.

Thanks a lot!
Amir
P.S: Attached: input file and a jpeg of this simple geometry

 

[amirbio]

and this is the jpeg.

 

[rstupplebeen]

Amir,
Try adding a constraint in the Y direction on a node that has a 0 Y coordinate. I believe that your are simply getting a rigid body mode due to numerical noise. Depending on your loading you may be able to only model the top half and constrain the cut. I hope this helps.


Rob Stupplebeen

https://sites.google.com/site/robertkstupplebeen/

 

[amirbio]

Dear Rstupplebeen,

Thanks for your reply. If I want to do that (substituting a 1/4 of the circle instead of a half circle) my first question is that does it still resemble the axisymmetric model of a sphere? if yes, what boundary conditions do I need to put on the lower end of the top 1/4 circle for example to reproduce the presence of the omitted lower section?
I would be very grateful if you could comment on this. It also reduces the calculation times and possibly I can get rid of that noise.

Many thanks,
Amir

 

[amirbio]

P.S: the node at 0,0 is the same as cavity reference node. This node, remains constant in its position through out the whole inflation process (I have checked this). Additionally, since this node is not a physical part of the instance, fixing this doesnt do anything to the ball movement. Is it possible to define a node that has a physical meaning and constraints can be applied to it but not located on a part? I mean a point to be pinned or fixed otherwise?
But the desired result would be that it doesnt move at al, since the load (inflation) is just symmetrical.
Thanks.